Shuffle Tracking

1357111317

Well-Known Member
#1
Even though this is my first post I have been following this site for a while. My question is, what kind of results can I expect with the following system and if the system I am using has major flaws please suggest alternatives:
( First the casino is pretty standard, 6 decks, pays 3:2, Hits on soft 17 and plays to about a deck left, max bet of 10 times the min bet)
The system I have been practicing and am resonably good at ( Not perfect, still hours of practice left. I can count a full deck in about 25 seconds though) is shuffle tracking.

What I do is count ( using Hi-Lo) every 26 cards and use a chip clock to keep track of it. Once the deck has been played through (roughly one deck left) I add up the total count and whatever the count of the last deck is I split it evenely into two half decks. Then once I have mapped the deck out I follow the shuffle.

*** Shuffle description removed by request ***

So that is basically the system I use. Fairly tough to use in full, 7 player tables but reasonably easy to pull off in 2-3 player tables. Thanks for any suggestions, comments or criticisms.
 

RJT

Well-Known Member
#2
1357111317 said:
Even though this is my first post I have been following this site for a while. My question is, what kind of results can I expect with the following system and if the system I am using has major flaws please suggest alternatives:
( First the casino is pretty standard, 6 decks, pays 3:2, Hits on soft 17 and plays to about a deck left, max bet of 10 times the min bet)
The system I have been practicing and am resonably good at ( Not perfect, still hours of practice left. I can count a full deck in about 25 seconds though) is shuffle tracking.

What I do is count ( using Hi-Lo) every 26 cards and use a chip clock to keep track of it. Once the deck has been played through (roughly one deck left) I add up the total count and whatever the count of the last deck is I split it evenely into two half decks. Then once I have mapped the deck out I follow the shuffle.

*** section removed by request ***

So that is basically the system I use. Fairly tough to use in full, 7 player tables but reasonably easy to pull off in 2-3 player tables. Thanks for any suggestions, comments or criticisms.
I'm afraid that you would require an unfeesably accurate dealer for this to work. You seem to have found a casino that uses single pass shuffles, but even at this - have you encountered a dealer that breaks the stack exactly into 2 156 card piles and then proceed to grab exactly 26 cards with each hand for each riffle? If he's one or two cards off on one, it might not make too much of a difference, but as he goes through the errors compound and your lovely little map telling you the count of each packet will put you enough out to destroy your bankroll.

RJT.
 
#3
I may have to re-read your post but it sounds like a shuffle that lends itself
to simpler techniques like cut-off tracking and/or tops and bottoms. zg
 
#4
RJT said:
I'm afraid that you would require an unfeesably accurate dealer for this to work. You seem to have found a casino that uses single pass shuffles, but even at this - have you encountered a dealer that breaks the stack exactly into 2 156 card piles and then proceed to grab exactly 26 cards with each hand for each riffle? If he's one or two cards off on one, it might not make too much of a difference, but as he goes through the errors compound and your lovely little map telling you the count of each packet will put you enough out to destroy your bankroll.

RJT.
Where the heck did you get that from? We're not looking for exact locations of cards with shuffle tracking, just a general idea of what the shoe is going to look like after the shuffle. With such a shuffle the dealer grabs can be way way off and the advantage is going to be way way higher than straight counting, and as long as he doesn't do anything silly his bankroll will be just fine.

The way I handle a shuffle like that is to break the daughter shoe down into a manageable number of segments, maybe 4, and rather than trying to micromanage the output just keep 4 running totals of contributions for each quarter shoe. That will give you a place to play and a place to avoid. Being this kind of tracking requires counting anyway, any info you get is additive to your counting.

Recommend to Numbers Guy that he not disclose the location of this shuffle nor give details about specific shuffles at all.
 

iCountNTrack

Well-Known Member
#5
, Number Man's shuffle is a simple Restack riffle which is ideal for brute force shuffle tracking to map the shoe. The smaller your tracking zone the more accurate is your shuffle tracking

there is no harm in describing the shuffle in detail as long as you don't release it location..
 

1357111317

Well-Known Member
#6
RJT said:
I'm afraid that you would require an unfeesably accurate dealer for this to work. You seem to have found a casino that uses single pass shuffles, but even at this - have you encountered a dealer that breaks the stack exactly into 2 156 card piles and then proceed to grab exactly 26 cards with each hand for each riffle? If he's one or two cards off on one, it might not make too much of a difference, but as he goes through the errors compound and your lovely little map telling you the count of each packet will put you enough out to destroy your bankroll.

RJT.
If you guys are wondering about the number I'm a math guy so those are the first 6 prime numbers haha. Well by observing the dealers They seem to be pretty accurate in their cuts. Cutting the two decks in half would be pretty accurate if you are doing it all day every day. And even if they are one or two cards off the first couple decks are going to be pretty accurate anyways right? And even in the lower decks won't the approximation be close enough to give me enough of an edge over the casino?
 

RJT

Well-Known Member
#7
Automatic Monkey said:
Where the heck did you get that from? We're not looking for exact locations of cards with shuffle tracking, just a general idea of what the shoe is going to look like after the shuffle. With such a shuffle the dealer grabs can be way way off and the advantage is going to be way way higher than straight counting, and as long as he doesn't do anything silly his bankroll will be just fine.

The way I handle a shuffle like that is to break the daughter shoe down into a manageable number of segments, maybe 4, and rather than trying to micromanage the output just keep 4 running totals of contributions for each quarter shoe. That will give you a place to play and a place to avoid. Being this kind of tracking requires counting anyway, any info you get is additive to your counting.

Recommend to Numbers Guy that he not disclose the location of this shuffle nor give details about specific shuffles at all.
You do whatever you want my good man - i think i'll stick with what i know works ;) Anyway, no point lecturing to an arrogant young whipper snapper like me. No doubt i'd think i know enough to ignore just about everything you post.

RJT.
 

RJT

Well-Known Member
#8
1357111317 said:
If you guys are wondering about the number I'm a math guy so those are the first 6 prime numbers haha. Well by observing the dealers They seem to be pretty accurate in their cuts. Cutting the two decks in half would be pretty accurate if you are doing it all day every day. And even if they are one or two cards off the first couple decks are going to be pretty accurate anyways right? And even in the lower decks won't the approximation be close enough to give me enough of an edge over the casino?
The reality is that even though most dealers seem accurate - they would need to be beyond realistically precise to use map tracking like this. No matter how accurate the dealer, if you don't use your eye to make adjustments, you ain't gonna win and very few people have the patience to train their eye to be as accurate as it needs to be.
Small errors shuffle tracking cost big bucks. If you are interested in shuffle tracking start off with Snyder's Cookbook then read alienated's post on the NRS formula over at the CCC - you can find the link in the thread just below this one. Neither of these will tell you everything you need to know, but they are a very good starting place. Before you do any of that though, make sure your counting is second nature.

RJT.
 

1357111317

Well-Known Member
#9
RJT said:
The reality is that even though most dealers seem accurate - they would need to be beyond realistically precise to use map tracking like this. No matter how accurate the dealer, if you don't use your eye to make adjustments, you ain't gonna win and very few people have the patience to train their eye to be as accurate as it needs to be.
Small errors shuffle tracking cost big bucks. If you are interested in shuffle tracking start off with Snyder's Cookbook then read alienated's post on the NRS formula over at the CCC - you can find the link in the thread just below this one. Neither of these will tell you everything you need to know, but they are a very good starting place. Before you do any of that though, make sure your counting is second nature.

RJT.
Now I had a quick look at those links and The NRS method seems awfully awfully complicated ( keep in mind I only had a quick look). Is this NRS method considered to be one of the most powerful systems out there? If not what are some of the most poweful systems?
 
#10
1357111317 said:
Now I had a quick look at those links and The NRS method seems awfully awfully complicated ( keep in mind I only had a quick look). Is this NRS method considered to be one of the most powerful systems out there? If not what are some of the most poweful systems?
What you are doing is very good and beyond what most people can do. Not every AP is a mentalist who is able to keep 12 different counts for each half deck throughout a shoe while playing, perform the shuffle transfer function on it, and get useful data for the next shoe but if you can do it, you can do it.

Not to get unnecessarily mathematical, but I did not use that phrase "transfer function" in vain. If you are mentally bored you can do a Google search on that phrase as well as "Fourier transform" to get a idea of the math used to solve a shuffle, and included in that math is the effect of the errors induced by dealer variability. (It's way worse math homework than NRS!)

To get a visual representation of this, make a bar graph that simulates a shoe- green bars going up to represent positive half-decks and red bars going down to represent negative half-decks. Nice picture, right? Now if you wear glasses, take them off, or if you don't put a pair of reading glasses on or do something to make your eyes go out of focus. What you are seeing is the effect of dealer grab variation and player deck estimation error. Not as clear, but you are still getting a lot of info from the graph. Back up, and the blurring gets worse, until you aren't getting any info from the graph at all. But once you get to that point, change the half-deck bars to full-deck, and you can see something again! Switch to two-deck bars, and you'll see where the best third and the worst third of the shoe is from quite a distance. Once you dig all this, you can come up with the best shuffle tracking procedure for your abilities and that particular shuffle.
 

1357111317

Well-Known Member
#11
Automatic Monkey said:
What you are doing is very good and beyond what most people can do. Not every AP is a mentalist who is able to keep 12 different counts for each half deck throughout a shoe while playing, perform the shuffle transfer function on it, and get useful data for the next shoe but if you can do it, you can do it.

Not to get unnecessarily mathematical, but I did not use that phrase "transfer function" in vain. If you are mentally bored you can do a Google search on that phrase as well as "Fourier transform" to get a idea of the math used to solve a shuffle, and included in that math is the effect of the errors induced by dealer variability. (It's way worse math homework than NRS!)

To get a visual representation of this, make a bar graph that simulates a shoe- green bars going up to represent positive half-decks and red bars going down to represent negative half-decks. Nice picture, right? Now if you wear glasses, take them off, or if you don't put a pair of reading glasses on or do something to make your eyes go out of focus. What you are seeing is the effect of dealer grab variation and player deck estimation error. Not as clear, but you are still getting a lot of info from the graph. Back up, and the blurring gets worse, until you aren't getting any info from the graph at all. But once you get to that point, change the half-deck bars to full-deck, and you can see something again! Switch to two-deck bars, and you'll see where the best third and the worst third of the shoe is from quite a distance. Once you dig all this, you can come up with the best shuffle tracking procedure for your abilities and that particular shuffle.
I appreciate the comments Automatic Monkey. To address your first comment I am not remember every single count, I am using chip clocks. Now I know this might seem a bit tough to pull off but i feel like I can camoflage it pretty well. ( On a side note using a chip clock isn't actually considered cheating is it? The casinos just treat it the same as card counting?)

I'm not sure what you are talking about peforming a shuffle transfer. In my method they take the two half decks and shuffle them together, then I just combine the counts of those two decks into one deck.

So in your opinion Monkey do you think this is a viable advantage method in the long term?
 

RJT

Well-Known Member
#12
Primes - the NRS formula isn't in itself a method for shuffle tracking, it is simply used to calculate the TC within a good packet and is actually far easier than it seems. A lot of it boils down to constants and if you have a few of the constants memorised for selected situation that you are likely to encounter its not actually that much more difficult to use than a TC calculation.
If you read the Cookbook, you'll learn the skills you need to be able to shuffle track sucessfully and why errors are so costly. Rather than trying to track the entire stack in the way you are suggesting you should be trying to track a specific packet of either good or bad cards through the shuffle and using this information to either cut a good packet into play (where you can bet high in the good packet) or cut the bad packet out of play (where you can bet highish through the whole shoe). And this is where your visual skills have to be up to par, so that you can ensure that the packet is where you think it is in the final stack and hasn't been split too much so as to have given weak or useless information.
Don't get me wrong, mapping the shuffle will give you some useful information. If you know roughly what packet will get mixed with the packet you are tracking you can take a count of this packet and use it to adjust your estimates. But just flat out map tracking in the way you're suggesting simply doesn't work.

RJT.
 
#13
RJT said:
Primes - the NRS formula isn't in itself a method for shuffle tracking, it is simply used to calculate the TC within a good packet and is actually far easier than it seems. A lot of it boils down to constants and if you have a few of the constants memorised for selected situation that you are likely to encounter its not actually that much more difficult to use than a TC calculation.
If you read the Cookbook, you'll learn the skills you need to be able to shuffle track sucessfully and why errors are so costly. Rather than trying to track the entire stack in the way you are suggesting you should be trying to track a specific packet of either good or bad cards through the shuffle and using this information to either cut a good packet into play (where you can bet high in the good packet) or cut the bad packet out of play (where you can bet highish through the whole shoe). And this is where your visual skills have to be up to par, so that you can ensure that the packet is where you think it is in the final stack and hasn't been split too much so as to have given weak or useless information.
Don't get me wrong, mapping the shuffle will give you some useful information. If you know roughly what packet will get mixed with the packet you are tracking you can take a count of this packet and use it to adjust your estimates. But just flat out map tracking in the way you're suggesting simply doesn't work.

RJT.
Not to turn you off to visual tracking methods, because with the kind of shuffle described any method would work.

But it would be very difficult for any method using a fully mapped shuffle to be as inaccurate as the visual estimation you recommend.
 

RJT

Well-Known Member
#14
So you've told us all before - it's great that you try to find ways to get an edge when your physical capibilities ain't up to the job.

RJT.

Edit - it might also be worth considering the merit of any technique that blindly counts on the dealer's skills rather than your own. If it seems too good to be true.....
 

1357111317

Well-Known Member
#15
So it looks like I have one person who thinks the method will work and one person who thinks it doesnt. Anyone else have any opinions?

Now Automatic do you think that it would work better counting every deck and then when it is all combined I am left with 3 different sections counted instead of 6? Would the decrease in precision be worth it because the shuffling errors make less of an impact when the sections I count are bigger? Or would the increased precision in the mapping be more benificial then the magnification of shuffling errors? Another thing I was thinking about was just mapping the aces. Would mapping the aces be more effective than the method I am using currently?
 

Sonny

Well-Known Member
#16
1357111317 said:
So it looks like I have one person who thinks the method will work and one person who thinks it doesnt.
I think we all agree that your technique will work in terms of profiling the shoe. The only disagreement is whether such a vague approach will give you an advantage over the casino or not.

Personally I think it’s a moot point. You haven’t mentioned anything about the most important skills: estimating and cutting. If you can’t accurately estimate the discards, estimate the grab sizes, estimate the lying stack and cut the stack (within 3-4 cards) then all of your tracking skill doesn’t matter. You've got to be able to cut to a segment and/or recognize it before it is dealt in order to make money. Acquiring those skills should be your first goal. Shuffle tracking is much more difficult than card counting.

My advice would be to practice your skills at home, then try tracking one or two segments in the casino. Once you build up your skill and confidence you can think about tracking more segments. You’re going to find that visually tracking the entire shoe by yourself is impossible to do accurately. There are just too many adjustments that have to be made. You either have to track a few segments accurately or track many segments vaguely. You can have quality or quantity. If it was my money, I’d want to bet on accuracy even though I would only get a few opportunities to play it.

There was thread similar to this one about 2 months ago. Here is a link with some good responses:

http://www.blackjackinfo.com/bb/showthread.php?t=10912

-Sonny-

P.S. - Don't bother mapping aces. Unless you know where they are going to land, the overall count will give you more information.
 
#17
1357111317 said:
So it looks like I have one person who thinks the method will work and one person who thinks it doesnt. Anyone else have any opinions?

Now Automatic do you think that it would work better counting every deck and then when it is all combined I am left with 3 different sections counted instead of 6? Would the decrease in precision be worth it because the shuffling errors make less of an impact when the sections I count are bigger? Or would the increased precision in the mapping be more benificial then the magnification of shuffling errors? Another thing I was thinking about was just mapping the aces. Would mapping the aces be more effective than the method I am using currently?
I think you have the right idea mathematically.

Let's assume irregularity of the dealer grab sizes and errors in your deck estimation are there. They're there, and there's nothing you can do about it.

Even so, shuffle tracking can give you such an enormous advantage the standard deviation of your estimated count can be huge and you'll still have a bigger advantage than with straight counting. With one shuffle I have tracked, a typical output for a segment would be a High-Low true count of +5, plus or minus 5. That might seem like an enormous spread, knowing that the true count will be somewhere between 0 and +10 about two-thirds of the time, but I'll put a max bet out on that all day!

To compensate for the errors (we've already stipulated they're unavoidable!) you have to do your homework and calculate what the effect of the shuffling and estimation irregularities are. You will need to use a computer and have some programming skill. What you will end up with, God willing, is a linear equation that you will be able to implement at the table, using the data from your segment counts and chip counting, but without using a computer. This will give you your average true count, and working on your deck estimation and finding dealers with grabs closest to optimum will reduce the standard deviation and cause you to see actual cards closest to what you have predicted.

The problem with merely following a pack of high cards through the shuffle is that the method implies complete ignorance of what cards that pack is mixed with. Thus with a single-pass shuffle, it can be no better than tracking a dealer who's grab varies by 50%.

Changing the resolution of your segment count makes it less controlled by irregularities but also washes away a lot of the information. So as long as you can handle all that data at the table I'd stay right with what you're doing now. The most important thing you have to work on, I'd say, is proper bet sizing. A lot of people have gotten in trouble with shuffle tracking by overbetting, because they are seeing huge counts every other shoe but not taking into account the variance introduced by irregularities in the shuffle.

Agree with Sonny: tracking aces is a very different game where it makes a difference exactly which hand the ace lands on. Different techniques.
 

iCountNTrack

Well-Known Member
#18
As Sonny has mentioned, you need to practice your deck estimations to within 2 or 3 cards before anything. Brute force shuffle tracking to map the whole is shoe is doable, but could be tedious. The most important thing if you were to track all the segments of the shoe is to have dealers who EXACTLY reproduce their shuffle.

You can generate some simple formulas that will map the shoe. for example a six deck shoe where the penetration is 75%, at the end of the shoe we will have the following:

i
h
g
f
e
d
c
b
a

where each lower case letter represents the count for of the half-deck slug that you tracked (there are 9 of them) and the order is the same as it would be in the discard tray bottom to top.

Capital K represents the count for the slug that got cut out of the play

Lets say the dealer plugs a slug K/3 every 1.5 decks, the shoe would look like

K/3
i
h
g
K/3
f
e
d
K/3
c
b
a

After some zone shuffle (too lazy to describe it in details) the new shoe would like the following

(C+K/3)/2
(C+K/3)/2
(A+G)/2
(A+G)/2
(D+F)/2
(D+F)/2
(B+H)/2
(B+H)/2
(K/3+I)/2
(K/3+I)/2
(E+K/3)/2
(E+K/3)/2

Not so hard, but again the REPRODUCIBILITY is super-important, because clearly if the dealer plugs the unplayed cards in different locations, or the ORDER of the grabs is changed, the above map would clearly change.

By the mini-lecture about the difference between accuracy and precision. Accuracy is how close your measurement is too the true value, precision is how consistent your measurements are, for shuffle-trackers a precise dealer is by far more important.

One last thing, probably stating the obvious if for instance the above formula gives a net count of -2 for one of the slugs, I wouldn't bet my life on it but a -6 or less would be a safe zone :).
 

1357111317

Well-Known Member
#19
iCountNTrack said:
As Sonny has mentioned, you need to practice your deck estimations to within 2 or 3 cards before anything. Brute force shuffle tracking to map the whole is shoe is doable, but could be tedious. The most important thing if you were to track all the segments of the shoe is to have dealers who EXACTLY reproduce their shuffle.

You can generate some simple formulas that will map the shoe. for example a six deck shoe where the penetration is 75%, at the end of the shoe we will have the following:

i
h
g
f
e
d
c
b
a

where each lower case letter represents the count for of the half-deck slug that you tracked (there are 9 of them) and the order is the same as it would be in the discard tray bottom to top.

Capital K represents the count for the slug that got cut out of the play

Lets say the dealer plugs a slug K/3 every 1.5 decks, the shoe would look like

K/3
i
h
g
K/3
f
e
d
K/3
c
b
a

After some zone shuffle (too lazy to describe it in details) the new shoe would like the following

(C+K/3)/2
(C+K/3)/2
(A+G)/2
(A+G)/2
(D+F)/2
(D+F)/2
(B+H)/2
(B+H)/2
(K/3+I)/2
(K/3+I)/2
(E+K/3)/2
(E+K/3)/2

Not so hard, but again the REPRODUCIBILITY is super-important, because clearly if the dealer plugs the unplayed cards in different locations, or the ORDER of the grabs is changed, the above map would clearly change.

By the mini-lecture about the difference between accuracy and precision. Accuracy is how close your measurement is too the true value, precision is how consistent your measurements are, for shuffle-trackers a precise dealer is by far more important.

One last thing, probably stating the obvious if for instance the above formula gives a net count of -2 for one of the slugs, I wouldn't bet my life on it but a -6 or less would be a safe zone :).

I appreciate your comments CountNTrack. A couple things though. At my casino most dealers ( I will try and only play these dealers) just take all the unplayed cards, put them ontop of the discard pile and then split that 6 deck stack into two and then use the shuffle method described in the first post. So i just split that deck that was unplayed ( The game is 83% pen usually) into two and whatever the count is i just assume its split evenly between the two. Not a perfect assumption but its the best I can do.


Also you and sonny talk about deck estimation. Why do I need to know deck estimation? I know that I can cut the right amount of decks off each time but for my system I just count every 26 cards in my head and thats how I know the count of every 26 cards.

When you say safe zone do you mean that you can assume the actual count of the deck is somewhere between a plus/minus of the count you think it is?
 
#20
iCountNTrack said:
...By the mini-lecture about the difference between accuracy and precision. Accuracy is how close your measurement is too the true value, precision is how consistent your measurements are, for shuffle-trackers a precise dealer is by far more important...
Can't agree with that. I'd rather know how close I am to the true count. In terms of shuffle tracking, presumably you are doing all your math outside the casino. If you get to the casino and the dealer is doing something consistently a bit different (let's say consuming a pile in 7 grabs instead of 6) you now have to go home and do your math again. But if he's doing his 6 grabs with a +/- 30% grab variation, you know that most of the cards are still where they are supposed to be most of the time, and when you predict a high count you know you are somewhere between a neutral count and a super-high count. I'll take it.

Keep in mind that in terms of playing blackjack it doesn't matter what the count of a shoe or a segment is- all that matters is what cards come out on the next round and we will never be certain of that.
 
Top